|
Hi Group.
I have designed a PCB around Hans' Nixie multimeter design :
The PCB works just nicely, just been mounted on my newly aquired switch mode power supply, but i am having problems with the tubes.
I am using IN8-2 tubes, and have grounded the decimal point on the middle one to make the denoter of XX.X volts.
The problem i have - the transformer i selected for my power supply board for the tubes etc. is only putting out 165V DC after rectification, and that seems a bit low. The lighted cathodes in the tube steal the current from the decimal dot, so it goes out sometimes...
Any good ideas of what i can do ? A capacitor-diode doubler circuit seems a bit overkill, since i will be getting a voltage that is too high...
Maybe i need to redesign the power supply PCB layout ?
I have a proposed PCB layout here:
I have thought about two small PCB transformers, back to back to get around 230V to the tubes instead. I just had a bunch of these nice PCB transformers laying around from an Natural Gas furnace, with a 170V (unloaded) secondary and a 18V one too.
I know that putting a resistor on each cathode on the nixie would possibly solve the problem, but im not going to make new PCB's again (already on revision 2, since i swapped all connections on the nixies (0 to 9 etc :-O - dangerous stuff when you design two pcb's apart, haha)
TIA.
// Per.
|
--- In NEONIXIE-L@..., "per.zapro" <per.zapro@...> wrote: I have designed a PCB around Hans' Nixie multimeter design :
The PCB works just nicely, just been mounted on my newly aquired switch mode power supply, but i am having problems with the tubes.
I am using IN8-2 tubes, and have grounded the decimal point on the middle one to make the denoter of XX.X volts.
The problem i have - the transformer i selected for my power supply board for the tubes etc. is only putting out 165V DC after rectification, and that seems a bit low. The lighted cathodes in the tube steal the current from the decimal dot, so it goes out sometimes...
Any good ideas of what i can do ? A capacitor-diode doubler circuit seems a bit overkill, since i will be getting a voltage that is too high...
Maybe i need to redesign the power supply PCB layout ?
I have a proposed PCB layout here:
I have thought about two small PCB transformers, back to back to get around 230V to the tubes instead. I just had a bunch of these nice PCB transformers laying around from an Natural Gas furnace, with a 170V (unloaded) secondary and a 18V one too.
<snip> Am I correct in thinking that this transformer with 170 V and 18 V secondaries is the one that you're using, but is not providing sufficient voltage? If so, you could wire the secondaries in series to get 188 VAC, which should give you a good 10-15 volts more DC under load after rectification and filtering. A.J.
|
"per.zapro" <per.zapro@...> wrote:
Hi Group.
I have designed a PCB around Hans' Nixie multimeter design :
I'm curious why Hans turned it into a direct drive circuit. The CA3162 already outputs its data in a multiplexed format, and the timing, from what I can get out of the datasheet, seems plenty slow enough to run nixies without any problem. If he went mux'd, he would have needed only one 74141, instead of 3. He wouldn't needed the 7475 latches, and the 7404 hex inverter. That's 6 extra chips. Instead the muxing would have needed 3 MPSA42s (or MMBTA42; SOT23), 3 MPSA92 (or SOT23 equiv), and 11 extra resistors (3 per digit, plus a common 2R divider). You connect the A42s in a 'common base' fashion, to deal with the phase of the anode drive outputs (MSD, NSD, & LSD), which are active low. Tie the emitter of each A42 to its corresponding output. Tie all the bases (of the A42s) to the tap of the divider. Uses 2 1Ks off of +5V to Gnd, so the tap is +2.5V. Tie the collectors of the A42s thru a resistor (220K-470K) to the base of its corresponding A92. Connect a 10K resistor (3x) across the base to emitter of each A92, to speed it up. Connect the emitter of all three A 92s to your nixie B+ (180V-200V). The collector of each A92 then ties to the corresponding anode resistor (in existing circuit) of each nixie tube. And there you have it. For newbies, and the old farts that have been programming too long, you have to remember with transistors (and tubes, too), there are three different hookups in your toolbox: (1) The 'common emitter', which most of us are familiar with, where the emitter is at a fixed point (usually gnd for NPNs, and a positive supply for PNPs), the base gets the input signal, sometimes thru a base resistor, and the collector is the output. It has both voltage gain and current gain, but the signal is inverted. (2) Emitter follower (aka Common collector) where the collector is tied to a fixed point, the base is still the input, but now the emitter is the output. Its called an 'emitter follower', for a reason. The emitter output follows input signal at the base, with NO voltage amplification. That means if you tie the collector to +12V or +180V, and the signal fed to the base goes from +1V to +6V, and back again, the emitter will follow the base, less the Vbe drop (~0.6V for silicon at 25C), from 0.4V to 5.4V, and back, as long as the collector voltage is over 7V or so, or a lot more , like +180V. An emitter follower, however, does have current gain, so the input need only supply 1mA, for a 100mA load on the output (given a beta of 100 for that transistor). Its basically at the heart of those common 3 pin regulators, like the 7805. (3) Finally, the common base configuration. This one isn't seen too often, unless you deal with RF circuits. It does still have its place in digital circuits. In this arrangement, the base is tied to a fixed point, and the signal is fed the emitter. The output comes out of the collector. This circuit has no current gain, but can have voltage gain, and that output signal is in phase (not inverted) from the input signal. It can save you from throwing in an extra inverter package. I also use a common base stage to read the K0 output (NDX or index) from my dekatron circuits, where the cathodes are all tied up to ~60V, but the uC needs a signal between 0V to no more than 5V. Also, the key to charlieplexing with transistors, is to combine the 'common base' with the 'common emitter' hookups, and get a matrix where you can get "n^2-n" outputs with 'n' I/O signals. That's 6 out with 3 I/O, 12 out with 4I/O, 20 out with 5I/O, ... . Of course, only one output ON at any one time.
|
"threeneurons" <threeneurons@...> wrote:
"per.zapro" <per.zapro@> wrote:
I have designed a PCB around Hans' Nixie multimeter design :
... The CA3162 already outputs its data in a multiplexed format,
Here's a drawing of my last rant: Feel free to try it out. This place seems to still have the chip:
|
On 8/27/10 11:00 AM, threeneurons wrote: "threeneurons"<threeneurons@...> wrote:
"per.zapro"<per.zapro@> wrote:
I have designed a PCB around Hans' Nixie multimeter design :
... The CA3162 already outputs its data in a multiplexed format,
Here's a drawing of my last rant:
That's a pretty simple circuit to get the job done. One question - what sets the 280 uA current in the MPSA42 collector to MPSA92 base connections? Shouldn't there be a big resistor there? The MPSA42 will be saturated and the MPSA92 base is one diode drop below 180V. -- David Forbes, Tucson AZ
|
That's a pretty simple circuit to get the job done.
One question - what sets the 280 uA current in the MPSA42
collector to MPSA92 base connections? Shouldn't there be a big
resistor there? The MPSA42 will be saturated and the MPSA92
base is one diode drop below 180V.
--
David Forbes, Tucson AZ
I did a little 'Muntz' to rid myself of an extra resistor per stage. As you notice all the bases, of all the MPSA42s are tied to a roughly fixed 3.4V. When one of, the 3, stages is ON, its emitter is brought to GND, but thru a 10K limiting resistor. So in reality, the one side of the resistor is at GND (0V), and the other tie to the A42s emitter, which is 0.6V (or there abouts) below that 3.4V, or ~2.8V. That's 2.8V across the 10K resistor, which translates to 280uA. That 280uA comes out of the emitter leg mostly from the collector side, and a bit thru the base (~2% with a beta of 50). The voltage from collector to emitter is almost the full 180V; Vce=180V. Power dissipation in that A42 is 280uA x 180V = 50mW. A MPSA42 can handle upto 350mW, and the MMBTA42 upto 250mW. 50mW is only a fraction of that value. We should be safe, even if those 350mW & 250mW numbers are derated, for the worst case operating temp. If that 10K resistor is ever shorted, the transistor will try to draw over 400mA (over 70W), which just might make it physically pop, if the nixie supply is up to it. Just a tad of extra info. You can make a simple constant current device by adding a resistor to the emitter leg of a transistor, where the other end of the resistor tied to GND (0V), or some other fixed voltage (V1). The base is then tied to another fixed voltage (V2). V2 should be such that the base-emitter is forward biased. In the case of and silicon NPN, V2 should be at least 0.6V more positive than V1. The set current (flowing into the collector) will roughly be: I=(V2-V1-0.6)/R This is exactly what we have with the A42s of the anode drive section of the circuit above. Of course, its not perfect. That 0.6V number varies with temperature. I believe its about 23mV per degree C, if those brain cells from 30 years ago are still alive. David already knows this. He's been pushing around them electrons for some time now. Or was he doing it before electrons formed out of the primordial soup of the big bang.
|
On 8/27/10 2:15 PM, threeneurons wrote:
That's a pretty simple circuit to get the job done.
One question - what sets the 280 uA current in the MPSA42
collector to MPSA92 base connections? Shouldn't there be a big
resistor there? The MPSA42 will be saturated and the MPSA92
base is one diode drop below 180V.
--
David Forbes, Tucson AZ
I did a little 'Muntz' to rid myself of an extra resistor per stage.
David already knows this. He's been pushing around them electrons for some time now.
Yeah, I figured you were doing something tricky with those emitter pull-downs, but I'm too busy working on the Burning Man packing stuff to do a proper circuit analysis. Thanks for the lucid description. -- David Forbes, Tucson AZ
|
--- In NEONIXIE-L@..., "A.J." <a.j.franzman@...> wrote: Am I correct in thinking that this transformer with 170 V and 18 V secondaries is the one that you're using, but is not providing sufficient voltage? If so, you could wire the secondaries in series to get 188 VAC, which should give you a good 10-15 volts more DC under load after rectification and filtering.
Yep. I tried connecting the two secondaries in series to obtain a higher voltage, but found out that the 170 volt secondary just won't cut the mustard - it sags a lot under load, so i scrapped the idea of using that transformer, and i am now looking at an smps soloution instead. // Per.
|
--- In NEONIXIE-L@..., "threeneurons" <threeneurons@...> wrote: I'm curious why Hans turned it into a direct drive circuit. The CA3162 already outputs its data in a multiplexed format, and the timing, from what I can get out of the datasheet, seems plenty slow enough to run nixies without any problem. If he went mux'd, he would have needed only one 74141, instead of 3. He wouldn't needed the 7475 latches, and the 7404 hex inverter. That's 6 extra chips.
Yep. I breadborded the circuit to death before going with Hans' schematic all the way. I simply could not get it to work! - I had problems with ghosting, strange noise issues and so on... Wiring up his schematic works wonderfull, and gives me stable readout, and no flickering. I already made the PCB's, but i am rev 1,1 now, due to some dumb mistakes. You can see the outcome here: I just need to make a board like this for the Amp section and then a power supply. At the moment i tried making the exact power supply with the MC34063A switcher, but i have problems with noise. I tried adding 20 uF MKT capacitance on the output of the switcher, but the LSB keeps jumping up and down 2-3 counts. For the FET i am using an IRFBC40A. // Per.
|
--- In NEONIXIE-L@..., "threeneurons" <threeneurons@...> wrote: Here's a drawing of my last rant:
Care to explain what that 1K resistor i circled in red, does ??? Feel free to try it out. This place seems to still have the chip: I am temped to try it out - just a big waste of time laying out that elaborate PCB for all that chips spaced close together.... I bought two of them at my local dealer, thay were pretty expensive, but putting high tech PIC's into old gear is not correct in my opinion :-) // Per.
|
"per.zapro" <per.zapro@...> wrote:
"threeneurons" <threeneurons@> wrote:
Here's a drawing of my last rant:
Care to explain what that 1K resistor i circled in red, does ???
// Per.
You mean the 1K off the 470 ohm. Its part of a divider, so the base of all the transistors (tied to it) see ~3.4V. I pick a voltage somewhere between Vcc (5V) and Gnd because I don't know how far the outputs swing. If the chip is CMOS, I don't need to worry, because it will swing far 0 to 5V, as long as its not loaded down too much. If its bipolar, then it may not go up all the way to +5V. Look at the specs for any real TTL chip (74xx, 74S, 74LS,). Same goes for NMOS. They are really asymmetrical. The datasheet only mentions the BCD outputs, but doesn't give specs on the 'digit drive' (MSD, NSD, LSD) outputs.
|
"per.zapro" <per.zapro@...> wrote:
Yep. I breadborded the circuit to death before
going with Hans' schematic all the way. I simply could
not get it to work! - I had problems with ghosting,
strange noise issues and so on...
Wiring up his schematic works wonderfull, and gives me
stable readout, and no flickering.
You can see the outcome here:
I just need to make a board like this for the Amp section
and then a power supply.
At the moment i tried making the exact power supply with
the MC34063A switcher, but i have problems with noise. ...
// Per.
I have a couple of those CA3162 chips on order. I'll play with them, to verify my version. The switcher Hans uses is the MKI version of the MC34063. Its of limited use. Here's a schematic of my MK1.5 Version: There are several improvements. (1) C1 in the feedback path adds stability to the circuit. Its gets rid of what you call noise. The FET drive is improved, so it now both actively pulls Up, and Down, the signal applied to the power FETs gate. The old version only pulls up, and has a passive pull down (a 330 ohm resistor). This greatly improves efficiency. The old circuit could only deliver ~10mA. The new circuit, is the one I sell on eBay, and can easily deliver 45mA. Finally, he probably used the IRF830, just because he had them in his junk box. Its a crappy FET. Use the ones mention in the schematic. Any of those is much better than the IRF830. The IRF740A has been the standard used in this group. As important as voltage and ON resistance, an important spec is the gate charge. Like the ON resistance, it should be as low as possible. Its the Power FETs 'dirty secret', since it means the impedance of gate goes down, as frequency goes up. That means you need a circuit to drive the FET, that can move current, exactly what they advertise a FET doesn't need. High impedance at DC (zero hertz) means nothing other than the marketing people just stroking themselves.
|
--- In NEONIXIE-L@..., "threeneurons" <threeneurons@...> wrote: I have a couple of those CA3162 chips on order. I'll play with them, to verify
my version. Cool, please let me know if you indeed get it to work - after 10 days of tinkering with a similar circuit i gave up, and laid out the PCB for all those Chips... The switcher Hans uses is the MKI version of the MC34063. Its of limited use.
Here's a schematic of my MK1.5 Version:
There are several improvements. (1) C1 in the feedback path adds stability to
the circuit. Its gets rid of what you call noise. Ah cool - i'm no expert at SMPS's i'll admit that, so i just do what ever i can to make it work. The FET drive is improved, so it now both actively pulls Up, and
Down, the signal applied to the power FETs gate. The old version
only pulls up, and has a passive pull down (a 330 ohm resistor).
This greatly improves efficiency. The old circuit could only
deliver ~10mA. The new circuit, is the one I sell on eBay, and can
easily deliver 45mA. Cool - i'll try to make it asap and see how it works. Finally, he probably used the IRF830, just because he had them in
his junk box. Its a crappy FET. Use the ones mention in the
schematic. Any of those is much better than the IRF830. The IRF740A
has been the standard used in this group. As important as voltage
and ON resistance, an important spec is the gate charge. Like the
ON resistance, it should be as low as possible. Its the Power
FETs 'dirty secret', since it means the impedance of gate goes
down, as frequency goes up. That means you need a circuit to drive
the FET, that can move current, exactly what they advertise a FET
doesn't need. High impedance at DC (zero hertz) means nothing other
than the marketing people just stroking themselves. Ah yeah i see it now - if the gate capacitance is too high, switching losses go trough the roof, and the smps controller can't fully charge the gate in the on-period, and that messes with the circuit. I just looked in my junk box, and the "best" i can come up with is IRF630, IRF640,IRFP560,IRFX32N500 etc. I think the best one of them is the IRF630, as mentioned on your schematic too, but 200 volts are a little low - what will happen when the circuit hits the 200 volts ? Does the transistor go into saturation, or will it simply breakdown and short out ? Another important component is of course the switching diode. In my test, i have been using an MUR1560, it seems that it does the job quite nicely - but are a little on the big side, mechanically. I guess the same appiles here - faster diode equals lower losses, and hence lower heat. // Per.
|
Here's a schematic of my MK1.5 Version:
I think the best one of them is the IRF630, as mentioned on
your schematic too, but 200 volts are a little low - what
will happen when the circuit hits the 200 volts ? Does the
transistor go into saturation, or will it simply breakdown
and short out ?
They don't actually breakdown right at 200V. Usually something rated at 200V may not breakdown until it hits close to 250V, so there is a bit of manufacturer margin built in. But that said, don't rely on it. You may get a batch that breakdown at 210V. If the total energy is limited, it may not be fatal. Power FETs are pretty tolerant with overvoltage across the drain to source. Overvoltage at the gate tends to be more destructive.
Another important component is of course the switching diode.
// Per.
All the components are important. The circuit is only as good as its weakest link. Even though, if there's one part that's of most importance, its the coil. That's where the magic is done. If you got a crappy coil, you've got nothing. That can be extended to all types on switching supplies. The magnetics (inductors & transformers) are key components the rest of the design is built around. That's why most professional switching supplies use custom magnetics.
|
